private void PreprocessInput()
{
FileInfo sourceSegModel =
this.customDir.GetFiles().Where(x => Regex.IsMatch(x.Name, "source.(spm|bpe)")).Single();
FileInfo targetSegModel =
this.customDir.GetFiles().Where(x => Regex.IsMatch(x.Name, "target.(spm|bpe)")).Single();
this.segmentationMethod = sourceSegModel.Extension;
var targetPrefix = this.model.TargetLanguages.Count > 1 ? this.targetLanguage.OriginalCode : null;
this.spSource = MarianHelper.PreprocessLanguage(
this.customSource,
this.customDir,
this.sourceLanguage.OriginalCode,
sourceSegModel,
this.includePlaceholderTags,
this.includeTagPairs,
targetPrefix);
this.spTarget = MarianHelper.PreprocessLanguage(
this.customTarget,
this.customDir,
this.targetLanguage.OriginalCode,
targetSegModel,
this.includePlaceholderTags,
this.includeTagPairs);
//concatenate the out-of-domain validation set with the in-domain validation set
ParallelFilePair tatoebaValidFileInfos =
HelperFunctions.GetTatoebaFileInfos(this.sourceLanguage.ShortestIsoCode, this.targetLanguage.ShortestIsoCode);
if (tatoebaValidFileInfos == null)
{
tatoebaValidFileInfos = HelperFunctions.GenerateDummyOODValidSet(this.customDir);
}
ParallelFilePair combinedValid = new ParallelFilePair(
tatoebaValidFileInfos,
new ParallelFilePair(this.inDomainValidationSource, this.inDomainValidationTarget),
this.customDir.FullName,
OpusCatMTEngineSettings.Default.OODValidSetSize);
this.spValidSource = MarianHelper.PreprocessLanguage(
combinedValid.Source,
this.customDir,
this.sourceLanguage.OriginalCode,
sourceSegModel,
this.includePlaceholderTags,
this.includeTagPairs,
targetPrefix);
this.spValidTarget = MarianHelper.PreprocessLanguage(
combinedValid.Target,
this.customDir,
this.targetLanguage.OriginalCode,
targetSegModel,
this.includePlaceholderTags,
this.includeTagPairs);
}
private void PreprocessInput()
{
var sourceSpm = this.customDir.GetFiles("source.spm").Single();
var targetSpm = this.customDir.GetFiles("target.spm").Single();
this.spSource = MarianHelper.PreprocessLanguage(this.customSource, this.customDir, this.sourceCode, sourceSpm, this.includePlaceholderTags, this.includeTagPairs);
this.spTarget = MarianHelper.PreprocessLanguage(this.customTarget, this.customDir, this.targetCode, targetSpm, this.includePlaceholderTags, this.includeTagPairs);
//concatenate the out-of-domain validation set with the in-domain validation set
ParallelFilePair tatoebaValidFileInfos = HelperFunctions.GetTatoebaFileInfos(this.sourceCode, this.targetCode);
ParallelFilePair combinedValid = new ParallelFilePair(
tatoebaValidFileInfos,
new ParallelFilePair(this.inDomainValidationSource, this.inDomainValidationTarget),
this.customDir.FullName,
OpusCatMTEngineSettings.Default.OODValidSetSize);
this.spValidSource = MarianHelper.PreprocessLanguage(combinedValid.Source, this.customDir, this.sourceCode, sourceSpm, this.includePlaceholderTags, this.includeTagPairs);
this.spValidTarget = MarianHelper.PreprocessLanguage(combinedValid.Target, this.customDir, this.targetCode, targetSpm, this.includePlaceholderTags, this.includeTagPairs);
}
